Why this poison is interesting

GHB is the only substance in Band C with a UK marketing authorisation for the molecule itself. Sodium oxybate is licensed for narcolepsy with cataplexy, and its summary of product characteristics states plainly that sodium oxybate is the sodium salt of GHB and that most of the overdose data derives from the illicit use of GHB.¹ That is a regulator-approved document describing a recreational drug's toxicology, and it makes this page unusually well sourced for a substance with no legitimate street form.

The second reason is the shape of the dose–response curve, and the label quantifies it: pharmacokinetics is nonlinear, with the area under the curve increasing 3.8-fold as the dose is doubled from 4.5 g to 9 g.¹ Doubling a dose and quadrupling the exposure is the arithmetic behind everything users and clinicians observe about this drug — that the difference between a recreational dose and unconsciousness is a matter of a millilitre or two, and that the same person can take the same volume on two nights with entirely different results.

The third is that GHB is the clearest case in the library of a poisoning where doing nothing is the treatment. The half-life is 0.5 to 1 hour¹, the drug is metabolised to carbon dioxide and breathed out¹, and every attempt to shorten the coma pharmacologically has failed. Zvosec and colleagues reviewed 18 published GHB toxicity case series and concluded that supportive care results in universally good outcomes.²

A poison is a drug whose kinetics have escaped its pharmacology.

The toxic principle

GHB is an endogenous compound in the mammalian brain, present at low concentrations, with both low- and high-affinity receptor targets.⁴ That is unusual in itself — it is one of very few recreational drugs that the brain already makes — and it is the reason its pharmacology took decades to disentangle.

GABA-B agonism accounts for the clinical picture directly: profound CNS depression, muscular hypotonia, bradycardia, hypothermia and respiratory depression. It is the same receptor baclofen acts at, which is why baclofen overdose can look strikingly similar — and why baclofen, unlike GHB, appears in the EXTRIP index.

GBL and 1,4-butanediol are prodrugs. GBL is converted to GHB by peripheral lactonase, and 1,4-butanediol by alcohol dehydrogenase and aldehyde dehydrogenase — the same enzymes that feature on the methanol and ethylene glycol pages. GBL is more lipid-soluble and more completely absorbed than GHB, so an equivalent volume delivers more drug and does so faster. Users switching between GHB and GBL on the basis of volume rather than content is a recognised route into an overdose, and it is one of the few dose errors in this library that has a purely pharmaceutical explanation.

Toxicokinetics

Unusually for a recreational drug, every row below except the last is taken from a UK summary of product characteristics for the molecule itself.¹ The table is the strongest documentary evidence on any page in Band C.

GHB — from the sodium oxybate label, for the molecule sold illicitly¹
Parameter	Therapeutic	In overdose	Why it changes
Absorption	Rapid and almost complete; absolute bioavailability about 88%. Tmax 0.5 to 2 hours across eight pharmacokinetic studies¹	Faster in practice — taken on an empty stomach in liquid form, effects begin within 15 minutes	Nothing delays this drug. There is no formulation, no enteric coating, no tablet to dissolve. The contrast with the modified-release trap on the calcium-channel blocker page could not be sharper.
Food effect	A high-fat meal increased Tmax from 0.75 h to 2.0 h, reduced Cmax by a mean of 58% and AUC by 37%¹	The same, and it is a genuine clinical modifier	A halved peak concentration from having eaten is a large effect for a food interaction, and it is part of why the same volume produces such different results on different nights. It is the label's own explanation for a phenomenon users describe.
Dose proportionality	Nonlinear. AUC increases 3.8-fold as the dose is doubled from 4.5 g to 9 g. Plasma levels increase more than proportionally with increasing dose. Single doses greater than 4.5 g have not been studied¹	Unstudied and unpredictable above the therapeutic range	The central row of this page. Doubling the dose nearly quadruples the exposure, and the label admits it has no data above 4.5 g. This is the arithmetic behind 'one capful too many', and it is the reason no threshold or dose figure appears anywhere on this page.
Protein binding	Less than 1% at concentrations from 3 to 300 µg/mL¹	Unchanged	Almost entirely unbound — the lowest protein binding in the library. On this row alone GHB looks like an ideal candidate for dialysis, which makes the label's conclusion about extracorporeal treatment more interesting rather than less.
Volume of distribution	190 to 384 mL/kg — a hydrophilic compound¹	Unchanged	About 15 to 30 litres in an adult, which is small. Combined with the binding, the physical chemistry says dialysable. What defeats dialysis here is not distribution but speed — see the final row.
Metabolism	The major elimination pathway — the label attributes this pathway detail to animal studies, with the human corroboration being the urinary-excretion figure below. GHB dehydrogenase converts it to succinic semialdehyde, then succinic semialdehyde dehydrogenase to succinic acid, which enters the Krebs cycle. An alternative route is beta-oxidation to acetyl-CoA. No active metabolites have been identified¹	Saturable — hence the non-linearity above	The poison is metabolised into an ordinary intermediate of the citric acid cycle. There is no toxic metabolite, nothing to accumulate and nothing to block. This is the cleanest detoxification in the library — the drug becomes carbon dioxide and water.
Elimination	Almost entirely by biotransformation to carbon dioxide, eliminated by expiration. Less than 5% appears unchanged in urine within 6 to 8 hours; faecal excretion negligible¹	Unchanged	A poison exhaled as carbon dioxide. The kidney has essentially no role — the label notes that no pharmacokinetic study in renal dysfunction was conducted because none would be expected to matter.¹
Half-life	0.5 to 1 hour¹	Prolonged somewhat by saturation, but still measured in hours	Shorter than anything else in this library that has a published figure. It is the reason the coma lifts on its own, the reason recovery is abrupt rather than gradual, and the reason no intervention has ever been shown to shorten it.
Dialysability	—	Not warranted. The use of haemodialysis and other forms of extracorporeal medicinal product removal have not been studied in sodium oxybate overdose, but has been reported in cases of acidosis due to GHB overdose. However, due to the rapid metabolism of sodium oxybate, these measures may not be warranted¹. EXTRIP has never addressed GHB³	The most instructive dialysability row in the library. Every physical property argues for dialysis: under 1% bound, 190–384 mL/kg of distribution, small and water-soluble.¹ And it is still not worth doing, because the patient's own metabolism is faster than the machine. Physical chemistry is necessary and not sufficient.

Metabolism and the metabolites

This is the shortest metabolism section in the library and the one with the happiest ending: GHB is metabolised into carbon dioxide and water, via intermediates that are ordinary constituents of central metabolism.¹ There is nothing to detoxify, no cofactor consumed and no metabolite to fear.

GHB — a poison that becomes carbon dioxide¹

GBL, or 1,4-butanediolProdrugs. GBL by lactonase; 1,4-butanediol by alcohol and aldehyde dehydrogenase — the enzymes of the toxic alcohol pages
GHBThe active species, and the sodium salt of the licensed medicine sodium oxybate¹
Low-affinity binding at GABA-B receptorsComa, hypotonia, bradycardia, respiratory depressionThe major pharmacological effects of exogenous GHB are mediated by GABA-B receptors⁴
High-affinity GHB binding sites — GABA-A subtypes among the candidatesPhysiological role, largely unresolvedKnown for over three decades; molecular identity only recently being uncovered⁴
Succinic semialdehydeSaturable — this is where the 3.8-fold non-linearity comes from¹
Succinic acidAn ordinary Krebs cycle intermediate. No active metabolites have been identified¹
Carbon dioxide and water, exhaledLess than 5% of the drug leaves unchanged in urine¹

One consequence of the metabolic route deserves separate mention because it produces a laboratory finding. The label records an increasing depth of coma ... at higher doses as well as acidosis, and notes events of hypernatraemia with metabolic alkalosis in the context of concomitant sodium chloride infusion.¹ The metabolic acidosis of a large GHB ingestion is a real and label-documented phenomenon; the alkalosis is iatrogenic, and the sodium load matters because the drug is administered and often sold as a sodium salt.

Elimination and accumulation

Nothing accumulates. The drug is exhaled as carbon dioxide within hours¹ and there is no active metabolite.¹ The accumulation that matters in GHB is neuroadaptive rather than chemical, and it produces the syndrome that is genuinely dangerous.

Withdrawal is the part of GHB that kills people in hospital

The Xyrem label notes that events associated with withdrawal syndrome have been observed outside the therapeutic range.¹ GHB is taken by dependent users at intervals of a few hours around the clock, because a half-life of 0.5 to 1 hour¹ means withdrawal begins within a few hours of the last dose. A dependent user admitted for any reason — an unrelated injury, an operation, a police custody episode — is in withdrawal by morning, and the syndrome is a severe autonomic and delirious state that is frequently refractory to ordinary benzodiazepine doses. The overdose recovers by itself in hours; the withdrawal does not. This is the single most important clinical point on the page and it is the one least often taught alongside the overdose.

The dosing interval is a direct kinetic consequence and is worth making explicit. Most drugs of dependence permit a daily or twice-daily rhythm. GHB's half-life forces a two- to four-hourly one, including overnight. A user who describes setting alarms through the night is describing dependence, and the history is diagnostic in a way that a volume or frequency estimate is not.

Why the recovery is abrupt rather than gradual

Most sedative poisonings resolve on a long tail: the patient becomes progressively rousable over many hours as the concentration falls slowly. GHB does not. A half-life of 0.5 to 1 hour¹ means the concentration falls fast, and the label describes consciousness that may fluctuate rapidly between a confusional, agitated combative state with ataxia and coma.¹ The patient who was unrousable twenty minutes ago sits up and pulls out their cannula — which is why the label warns that although the gag reflex may be absent in deeply comatose patients, even unconscious patients may become combative to intubation, and recommends rapid sequence induction without a sedative be considered.¹ A label advising RSI without sedation is describing a very particular kind of patient.

Target organs — and why those

GHB damages very little. That is the finding, not an omission — a drug that becomes carbon dioxide within hours and has no active metabolite has almost no opportunity for organ toxicity. Every card below except the last describes a functional depression that reverses completely.

Brain — cortical and reticular activating system

TargetGABA-B receptors, at which GHB binds with low affinity

Why hereThe dominant effect, and the reason a recreational dose rather than an endogenous concentration is required to produce it: the major pharmacological effects of exogenous GHB are mediated by GABA-B receptors, which bind GHB with low affinity, while the high-affinity sites — possibly certain GABA-A subtypes — are occupied at physiological concentrations.⁴ The drug's toxicity and the compound's physiology are pharmacologically separate events at separate receptors. Established

At the bedsideComa, fluctuating rapidly between a confusional, agitated combative state with ataxia and coma¹, with tendon reflexes intact¹. Myoclonus and tonic-clonic seizures are reported.¹ The fluctuation is the diagnostic feature.

Brainstem respiratory centres

TargetGABA-B-mediated depression of respiratory drive

Why hereThe mechanism that makes an otherwise self-limiting poisoning lethal. The label is explicit: reports of compromise in the rate and depth of respiration and of life-threatening respiratory depression, necessitating intubation and ventilation. Cheyne-Stokes respiration and apnoea have been observed.¹ The drug does not injure the brainstem; it silences it, and the patient dies of the interval rather than of the damage. Established

At the bedsideHypoventilation, Cheyne-Stokes breathing, apnoea.¹ This is the entire reason a GHB patient needs a hospital: the airway and the breathing must be carried for a few hours until the drug is exhaled.

Heart and thermoregulation

TargetReduced central sympathetic outflow

Why hereBradycardia and hypothermia accompany unconsciousness on the label¹, and both follow from central depression rather than from any direct cardiac or hypothalamic toxicity. The bradycardia has been responsive to intravenous atropine¹ — a small but useful detail, since it identifies the mechanism as vagally mediated rather than as conduction system disease. Established

At the bedsideBradycardia responsive to atropine¹; hypothermia. The bradycardia is usually well tolerated in a supine, sedated patient and rarely needs treating.

Airway — by vomiting rather than by pharmacology

TargetEmesis in the presence of impaired consciousness

Why hereThe commonest actual cause of harm in GHB toxicity, and it is mechanical. The label states that emesis (even with impaired consciousness) occurs, and directs that because emesis may occur in the presence of impaired consciousness, appropriate posture (left lateral recumbent position) and protection of the airway by intubation may be warranted.¹ A drug that causes vomiting and unconsciousness simultaneously is a drug that causes aspiration. Established

At the bedsideVomiting during coma; aspiration pneumonitis. The left lateral position is on the label for a reason and is the single most useful thing done for these patients before hospital.

Acid–base and sodium balance

TargetMetabolic consequences of a large ingestion, and of the sodium salt

Why hereTwo separate findings, one from the drug and one from its treatment. Acidosis is described at higher doses¹ — plausibly from the metabolic load and from hypoventilation, though the label does not attribute a mechanism. Hypernatraemia with metabolic alkalosis has been reported in the context of concomitant sodium chloride infusion¹, which is an interaction between a sodium-salt drug and a sodium-containing fluid rather than a toxic effect. Inferred

At the bedsideCheck gases and sodium in a deep or prolonged coma. The alkalosis is the one to notice, because it is caused by treatment rather than by the drug.

Timeline of effects

GHB — the library's fastest poisoning, start to finish

Time

What you seeWhat is happening

0–15 minOnset
What you seeEuphoria, disinhibition, then drowsiness.
What is happeningRapid and almost complete absorption, bioavailability about 88%, Tmax 0.5 to 2 hours by the label¹ — faster in practice on an empty stomach in liquid form. GBL is converted to GHB peripherally within minutes and is absorbed faster still.
15–60 minComa
What you seeDeep unconsciousness with intact tendon reflexes, hypotonia, bradycardia, hypothermia. Vomiting may occur while unconscious.¹
What is happeningLow-affinity GABA-B agonism.¹⁴ The steepness matters here: the label's 3.8-fold rise in AUC for a doubled dose¹ is why a small extra volume produces coma rather than sedation.
Any pointRespiratory depression
What you seeCheyne-Stokes respiration, apnoea, life-threatening respiratory depression necessitating intubation and ventilation.¹
What is happeningThe only mechanism on this timeline that kills. It does not follow a predictable point on the curve — it depends on dose, on co-ingestants, and above all on whether anyone is watching.
1–3 hMetabolism, and the fall
What you seeDepth of coma begins to lift.
What is happeningHalf-life 0.5 to 1 hour¹, with saturation of GHB dehydrogenase slowing clearance at the top of the curve. The drug is being converted to succinic acid and exhaled as carbon dioxide¹ — no antidote could accelerate this.
2–6 hAbrupt emergence
What you seeThe patient wakes suddenly, often agitated and combative, and frequently wants to leave.
What is happeningConsciousness may fluctuate rapidly between a confusional, agitated combative state with ataxia and coma¹. The label warns that even unconscious patients may become combative to intubation.¹ The abruptness is a kinetic consequence of the short half-life, not a behavioural one.
2–6 h after the last doseWithdrawal, in a dependent user
What you seeThe gap drawn here is the interval before withdrawal begins in someone dependent — tremor, tachycardia, hypertension, insomnia, then a severe agitated delirium.
What is happeningA half-life of 0.5 to 1 hour¹ means a dependent user redoses every two to four hours around the clock. Admission for any reason interrupts that, and the label notes withdrawal events observed outside the therapeutic range.¹ This phase, not the overdose, is where GHB patients deteriorate in hospital.
DaysWithdrawal delirium
What you seeA prolonged, benzodiazepine-refractory delirium requiring critical care in severe cases.
What is happeningNeuroadaptation to chronic GABA-B agonism, unmasked. The mirror image of the overdose, and the dangerous half of this drug.

What the mechanism predicts at the bedside

Deep coma with intact tendon reflexes should raise GHB.¹ It is not typical of the sedative-hypnotics generally and is one of the few genuinely discriminating examination findings in this band.
Fluctuation between combative agitation and coma is characteristic, and is described on the label in those words.¹
Left lateral position, on the label's own instruction.¹ Emesis in the presence of impaired consciousness is the mechanism most likely to cause lasting harm.
Expect abrupt emergence at two to six hours. The label warns that even unconscious patients may become combative to intubation and that RSI without a sedative should be considered.¹
Flumazenil will do nothing — no reversal of the central depressant effects ... can be expected from flumazenil¹ — and naloxone has insufficient evidence to be recommended.¹ A failure to respond to either is expected, not informative.
Ask when the last dose was, and whether they use through the night. A two- to four-hourly around-the-clock pattern means the patient will be in withdrawal before morning.
The withdrawal is the emergency, not the overdose. Plan for it at the point of admission rather than at the point it declares itself.
Ask about stimulant co-ingestion. GHB/GBL was co-used in 54.2% of 850 central London methamphetamine presentations, and that group needed more critical care⁵ — see amphetamines and MDMA.
Check gases and sodium in a deep coma: acidosis is described at higher doses, and hypernatraemia with alkalosis has followed sodium chloride infusion in a patient given a sodium salt.¹
Do not request dialysis. Every physical property suggests it would work; the label says the rapid metabolism means it may not be warranted¹, and EXTRIP has never addressed it.³

The antidote, from the poison's side

There is no antidote, and the label rules out the two candidates a clinician would reach for first. That is unusually direct for a regulatory document, and it is worth quoting exactly rather than paraphrasing.

Physostigmine — a proposed antidote that was tried and failed

Physostigmine was proposed for GHB coma on the basis of reported awakenings in patients anaesthetised with GHB and in five of six patients given it for GHB intoxication. Zvosec and colleagues examined what happened when a physostigmine protocol was actually instituted in an urban emergency department and ambulance service. Five cases of GHB toxicity received physostigmine. None demonstrated a response, and there were associated adverse events — atrial fibrillation in two, pulmonary infiltrates in one, significant bradycardia in one and hypotension in one. Reviewing 18 published GHB case series alongside, they conclude that physostigmine is not indicated for reversal of GHB-induced alteration of consciousness; it is not efficacious, it may be unsafe, particularly in the setting of recreational polydrug use; and supportive care results in universally good outcomes.² All five of their cases had co-intoxicants, which the authors themselves flag as the setting in which physostigmine's risk is heightened. Traditional teaching

Airway protection and ventilation: The treatment. The label describes life-threatening respiratory depression necessitating intubation and ventilation, Cheyne-Stokes respiration and apnoea¹ — and everything else on this page is a distraction from the fact that a patient whose breathing is supported for a few hours recovers.
Left lateral recumbent positioning: On the label, for the specific reason that emesis occurs in the presence of impaired consciousness.¹ The lowest-technology intervention in this library and among the highest-yield.
Atropine: For bradycardia, which the label records has been responsive to intravenous administration.¹ Rarely needed.
Flumazenil: No. Wrong receptor family — the label states no reversal can be expected.¹
Naloxone: No. Insufficient evidence to recommend¹. Its absence of effect is expected and does not argue against a GHB diagnosis.
Physostigmine: No. Five patients treated under a protocol, none responded, with five adverse events among them.²
Gastric lavage: May be considered if co-ingestants are suspected¹ — a narrow indication, and note what it implies: not for the GHB, which is absorbed within minutes, but for whatever else was taken.
Extracorporeal removal: Not warranted. Reported in cases of acidosis due to GHB overdose, but due to the rapid metabolism ... these measures may not be warranted.¹ EXTRIP has published nothing.³

Critical appraisal

The label describes sodium oxybate, given at therapeutic doses, and says so. It states that information about overdose is limited and that most data derives from the illicit use of GHB.¹ The overdose section is therefore itself a summary of the illicit literature rather than a trial dataset — which is exactly how it is used here, and is why the clinical descriptions are quoted rather than counted.
The 3.8-fold non-linearity figure is between 4.5 g and 9 g, and the label adds that single doses greater than 4.5 g have not been studied.¹ Extrapolating the ratio to recreational overdose doses is not supported by the source, and this page does not do it — the figure is used to establish that the curve is steep, not how steep it becomes.
The physostigmine badge rests on a study that directly tested the claim and refuted it. Zvosec and colleagues found five treated patients, no responders, and five adverse events among them, and state the conclusion in their own words.² That is a citation against the claim rather than an absence of one for it, which is what this library's house rule requires. The counter-argument is stated on the page: the original support was five of six patients responding, and Zvosec's five cases all had co-intoxicants — a small study on both sides.
The flumazenil and naloxone statements are the label's, and the mechanistic explanation for them is this page's. That flumazenil cannot work because GHB acts at GABA-B rather than at the benzodiazepine site of GABA-A follows from Bay's account of the receptor pharmacology⁴ and is a reconstruction rather than a quoted claim.
The receptor pharmacology is a research review, and it describes an unfinished field. Bay and colleagues state that the molecular identity of the high-affinity binding sites has only recently begun to be uncovered, and that certain GABA-A subtypes have emerged as candidates.⁴ The low-affinity GABA-B account of the exogenous drug's effects is secure; the high-affinity story is explicitly work in progress, and this page marks that rather than presenting it as settled.
The withdrawal syndrome is the most clinically important claim on this page and the least well evidenced here. The label records only that withdrawal events have been observed outside the therapeutic range.¹ The severity, the benzodiazepine refractoriness and the dosing-interval account are drawn from clinical description rather than from a cited study, and this page does not attach a frequency, a dose or a treatment regimen to any of them.
The 1,4-butanediol and ethanol competition is inference from shared enzymology, marked as such. Alcohol dehydrogenase handles both, exactly as on the methanol page; that the clinical consequence is a slower and longer conversion rather than a protective one is reasoning, not a finding.
The acidosis mechanism is not stated on the label, which reports the finding without explaining it.¹ The suggestion here that it reflects metabolic load and hypoventilation is inference and is flagged on the organ card.
No toxic dose, street volume, concentration or lethality figure appears on this page — the only doses given are the therapeutic ones the label used to establish the non-linearity (4.5 g and 9 g). Given a drug sold by the capful in solutions of unknown strength, nothing else could be given responsibly. Zvosec's other well-known paper is a series of GHB-associated deaths; it is deliberately not cited here, because its subject is lethality and this site does not print comparative lethality.
EXTRIP has never addressed GHB.³ Unlike clonidine, the absence leaves no open question: the label answers it, and the answer turns on speed rather than on physical chemistry.

References

1
Xyrem 500 mg/ml oral solution (sodium oxybate) — Summary of Product Characteristics. electronic medicines compendium, product 178. Sections 4.9 (Overdose) and 5.2 (Pharmacokinetic properties). States that sodium oxybate is the sodium salt of GHB and that most overdose data derives from the illicit use of GHB; source of the 3.8-fold AUC increase on doubling the dose from 4.5 g to 9 g, the 0.5–1 hour half-life, 88% bioavailability, the high-fat-meal effect, protein binding under 1%, volume of distribution 190–384 mL/kg, the metabolic pathway to carbon dioxide, the intact tendon reflexes, and the statements on flumazenil, naloxone and extracorporeal removal. medicines.org.uk/emc/product/178
2
Zvosec DL, Smith SW, Litonjua R, et al. Physostigmine for gamma-hydroxybutyrate coma: inefficacy, adverse events, and review. Clinical Toxicology 2007;45(3):261–5. PMID 17453877. (Five cases of GHB toxicity, all with co-intoxicants, received physostigmine under a departmental protocol; none demonstrated a response, with associated adverse events including atrial fibrillation in two, pulmonary infiltrates in one, significant bradycardia in one and hypotension in one. Reviewed 18 published GHB toxicity case series; concludes physostigmine is not indicated, is not efficacious, may be unsafe particularly in recreational polydrug use, and that supportive care results in universally good outcomes.)
3
EXTRIP Workgroup — published recommendations index. Extracorporeal Treatments in Poisoning Workgroup. Cited for the absence of any recommendation covering GHB, GBL or sodium oxybate. extrip-workgroup.org/recommendations
4
Bay T, Eghorn LF, Klein AB, et al. GHB receptor targets in the CNS: focus on high-affinity binding sites. Biochemical Pharmacology 2014 Jan 15;87(2):220–8. PMID 24269284. doi:10.1016/j.bcp.2013.10.028. (States that the major pharmacological effects of exogenous GHB are mediated by GABA-B receptors, which bind GHB with low affinity; high-affinity binding sites have been known for over three decades and their molecular identity has only recently begun to be uncovered, with certain GABA-A receptor subtypes emerging as candidates.)
5
Harnett JT, Dargan PI, Dines AM, et al. Increasing emergency department attendances in central London with methamphetamine toxicity and associated harms. Emergency Medicine Journal 2022 Jun;39(6):463–66. PMID 34649939. (GHB/GBL co-used in 54.2% of an 850-case methamphetamine series, associated with a higher Poisoning Severity Score and requirement for level 2/3 care.)

GHB and GBL

At a glance